mbox series

[0/3] mmc: sdhci-of-dwcmshc: support Sophgo SG2042

Message ID cover.1713258948.git.unicorn_wang@outlook.com
Headers show
Series mmc: sdhci-of-dwcmshc: support Sophgo SG2042 | expand

Message

Chen Wang April 16, 2024, 9:50 a.m. UTC
From: Chen Wang <unicorn_wang@outlook.com>

Add support for the mmc controller for Sophgo SG2042.
Adding corresponding new compatible strings, and implement
custom sdhci_ops.

This patchset is based on v6.9-rc1 and depends on following pathsets:

- [PATCH 0/1] mmc: sdhci-of-dwcmshc: enhance framework [1]
- [PATCH v14 0/5] riscv: sophgo: add clock support for sg2042 [2]
- [PATCH v1] mmc: sdhci-of-dwcmshc: th1520: Increase tuning loop count to 128 [3]


Link: https://lore.kernel.org/linux-kernel/cover.1713257181.git.unicorn_wang@outlook.com/ [1]
Link: https://lore.kernel.org/linux-riscv/cover.1713164546.git.unicorn_wang@outlook.com/ [2]
Link: https://lore.kernel.org/all/20240402093539.184287-1-bigunclemax@gmail.com/ [3]


---

Chen Wang (3):
  dt-bindings: mmc: sdhci-of-dwcmhsc: Add Sophgo SG2042 support
  mmc: sdhci-of-dwcmshc: Add support for Sophgo SG2042
  riscv: dts: add mmc controllers for Sophgo SG2042 SoC

 .../bindings/mmc/snps,dwcmshc-sdhci.yaml      |  67 +++++--
 .../boot/dts/sophgo/sg2042-milkv-pioneer.dts  |  15 ++
 arch/riscv/boot/dts/sophgo/sg2042.dtsi        |  32 ++++
 drivers/mmc/host/sdhci-of-dwcmshc.c           | 173 +++++++++++++++++-
 4 files changed, 264 insertions(+), 23 deletions(-)

Comments

Jisheng Zhang April 16, 2024, 3:37 p.m. UTC | #1
On Tue, Apr 16, 2024 at 05:50:57PM +0800, Chen Wang wrote:
> From: Chen Wang <unicorn_wang@outlook.com>
> 
> Add support for the mmc controller of Sophgo SG2042.
> 
> SG2042 uses Synopsys PHY the same as TH1520 so we reuse the tuning
> logic from TH1520. Besides this, this patch implement some SG2042
> specific work, such as clocks and reset ops.
> 
> Signed-off-by: Chen Wang <unicorn_wang@outlook.com>
> ---
>  drivers/mmc/host/sdhci-of-dwcmshc.c | 173 ++++++++++++++++++++++++++--
>  1 file changed, 166 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/mmc/host/sdhci-of-dwcmshc.c b/drivers/mmc/host/sdhci-of-dwcmshc.c
> index 1a0b7ded7f9f..432ce0398163 100644
> --- a/drivers/mmc/host/sdhci-of-dwcmshc.c
> +++ b/drivers/mmc/host/sdhci-of-dwcmshc.c
> @@ -106,12 +106,13 @@
>  #define DWC_MSHC_PTR_PHY_R	0x300
>  
>  /* PHY general configuration */
> -#define PHY_CNFG_R		(DWC_MSHC_PTR_PHY_R + 0x00)
> -#define PHY_CNFG_RSTN_DEASSERT	0x1  /* Deassert PHY reset */
> -#define PHY_CNFG_PAD_SP_MASK	GENMASK(19, 16) /* bits [19:16] */
> -#define PHY_CNFG_PAD_SP		0x0c /* PMOS TX drive strength */
> -#define PHY_CNFG_PAD_SN_MASK	GENMASK(23, 20) /* bits [23:20] */
> -#define PHY_CNFG_PAD_SN		0x0c /* NMOS TX drive strength */
> +#define PHY_CNFG_R			(DWC_MSHC_PTR_PHY_R + 0x00)
> +#define PHY_CNFG_RSTN_DEASSERT		0x1  /* Deassert PHY reset */
> +#define PHY_CNFG_PHY_PWRGOOD_MASK	BIT_MASK(1) /* bit [1] */
> +#define PHY_CNFG_PAD_SP_MASK		GENMASK(19, 16) /* bits [19:16] */
> +#define PHY_CNFG_PAD_SP			0x0c /* PMOS TX drive strength */
> +#define PHY_CNFG_PAD_SN_MASK		GENMASK(23, 20) /* bits [23:20] */
> +#define PHY_CNFG_PAD_SN			0x0c /* NMOS TX drive strength */
>  
>  /* PHY command/response pad settings */
>  #define PHY_CMDPAD_CNFG_R	(DWC_MSHC_PTR_PHY_R + 0x04)
> @@ -143,7 +144,8 @@
>  
>  /* PHY CLK delay line settings */
>  #define PHY_SDCLKDL_CNFG_R		(DWC_MSHC_PTR_PHY_R + 0x1d)
> -#define PHY_SDCLKDL_CNFG_UPDATE	BIT(4) /* set before writing to SDCLKDL_DC */
> +#define PHY_SDCLKDL_CNFG_EXTDLY_EN	BIT(0)
> +#define PHY_SDCLKDL_CNFG_UPDATE		BIT(4) /* set before writing to SDCLKDL_DC */
>  
>  /* PHY CLK delay line delay code */
>  #define PHY_SDCLKDL_DC_R		(DWC_MSHC_PTR_PHY_R + 0x1e)
> @@ -151,6 +153,9 @@
>  #define PHY_SDCLKDL_DC_DEFAULT		0x32 /* default delay code */
>  #define PHY_SDCLKDL_DC_HS400		0x18 /* delay code for HS400 mode */
>  
> +#define PHY_SMPLDL_CNFG_R		(DWC_MSHC_PTR_PHY_R + 0x20)
> +#define PHY_SMPLDL_CNFG_BYPASS_EN	BIT(1)
> +
>  /* PHY drift_cclk_rx delay line configuration setting */
>  #define PHY_ATDL_CNFG_R			(DWC_MSHC_PTR_PHY_R + 0x21)
>  #define PHY_ATDL_CNFG_INPSEL_MASK	GENMASK(3, 2) /* bits [3:2] */
> @@ -194,6 +199,11 @@ struct rk35xx_priv {
>  	u8 txclk_tapnum;
>  };
>  
> +#define SG2042_MAX_CLKS 2

I don't think "bulk" is suitable here for max 2 clks, no?

> +struct sg2042_priv {
> +	struct clk_bulk_data clks[SG2042_MAX_CLKS];

useless either

> +};
> +
>  struct dwcmshc_priv {
>  	struct clk	*bus_clk;
>  	int vendor_specific_area1; /* P_VENDOR_SPECIFIC_AREA reg */
> @@ -690,6 +700,76 @@ static void cv18xx_sdhci_reset(struct sdhci_host *host, u8 mask)
>  	sdhci_writel(host, val, priv->vendor_specific_area1 + CV18XX_SDHCI_PHY_TX_RX_DLY);
>  }
>  
> +static inline void sg2042_sdhci_phy_init(struct sdhci_host *host)
> +{
> +	u32 val;
> +
> +	/* Asset phy reset & set tx drive strength */
> +	val = sdhci_readl(host, PHY_CNFG_R);
> +	val &= ~PHY_CNFG_RSTN_DEASSERT;
> +	val |= FIELD_PREP(PHY_CNFG_PHY_PWRGOOD_MASK, 1);
> +	val |= FIELD_PREP(PHY_CNFG_PAD_SP_MASK, 9);
> +	val |= FIELD_PREP(PHY_CNFG_PAD_SN_MASK, 8);
> +	sdhci_writel(host, val, PHY_CNFG_R);
> +
> +	/* Configure phy pads */
> +	val = PHY_PAD_RXSEL_3V3;
> +	val |= FIELD_PREP(PHY_PAD_WEAKPULL_MASK, PHY_PAD_WEAKPULL_PULLUP);
> +	val |= FIELD_PREP(PHY_PAD_TXSLEW_CTRL_P_MASK, 3);
> +	val |= FIELD_PREP(PHY_PAD_TXSLEW_CTRL_N_MASK, 2);
> +	sdhci_writew(host, val, PHY_CMDPAD_CNFG_R);
> +	sdhci_writew(host, val, PHY_DATAPAD_CNFG_R);
> +	sdhci_writew(host, val, PHY_RSTNPAD_CNFG_R);
> +
> +	val = PHY_PAD_RXSEL_3V3;
> +	val |= FIELD_PREP(PHY_PAD_TXSLEW_CTRL_P_MASK, 3);
> +	val |= FIELD_PREP(PHY_PAD_TXSLEW_CTRL_N_MASK, 2);
> +	sdhci_writew(host, val, PHY_CLKPAD_CNFG_R);
> +
> +	val = PHY_PAD_RXSEL_3V3;
> +	val |= FIELD_PREP(PHY_PAD_WEAKPULL_MASK, PHY_PAD_WEAKPULL_PULLDOWN);
> +	val |= FIELD_PREP(PHY_PAD_TXSLEW_CTRL_P_MASK, 3);
> +	val |= FIELD_PREP(PHY_PAD_TXSLEW_CTRL_N_MASK, 2);
> +	sdhci_writew(host, val, PHY_STBPAD_CNFG_R);
> +
> +	/* Configure delay line */
> +	/* Enable fixed delay */
> +	sdhci_writeb(host, PHY_SDCLKDL_CNFG_EXTDLY_EN, PHY_SDCLKDL_CNFG_R);
> +	/*
> +	 * Set delay line.
> +	 * Its recommended that bit UPDATE_DC[4] is 1 when SDCLKDL_DC is being written.
> +	 * Ensure UPDATE_DC[4] is '0' when not updating code.
> +	 */
> +	val = sdhci_readb(host, PHY_SDCLKDL_CNFG_R);
> +	val |= PHY_SDCLKDL_CNFG_UPDATE;
> +	sdhci_writeb(host, val, PHY_SDCLKDL_CNFG_R);
> +	/* Add 10 * 70ps = 0.7ns for output delay */
> +	sdhci_writeb(host, 10, PHY_SDCLKDL_DC_R);
> +	val = sdhci_readb(host, PHY_SDCLKDL_CNFG_R);
> +	val &= ~(PHY_SDCLKDL_CNFG_UPDATE);
> +	sdhci_writeb(host, val, PHY_SDCLKDL_CNFG_R);
> +
> +	/* Set SMPLDL_CNFG, Bypass */
> +	sdhci_writeb(host, PHY_SMPLDL_CNFG_BYPASS_EN, PHY_SMPLDL_CNFG_R);
> +
> +	/* Set ATDL_CNFG, tuning clk not use for init */
> +	val = FIELD_PREP(PHY_ATDL_CNFG_INPSEL_MASK, 2);

magic "2" needs a meaningful macro definition.

> +	sdhci_writeb(host, val, PHY_ATDL_CNFG_R);
> +
> +	/* Deasset phy reset */
> +	val = sdhci_readl(host, PHY_CNFG_R);
> +	val |= PHY_CNFG_RSTN_DEASSERT;
> +	sdhci_writel(host, val, PHY_CNFG_R);
> +}
> +
> +static void sg2042_sdhci_reset(struct sdhci_host *host, u8 mask)
> +{
> +	sdhci_reset(host, mask);
> +
> +	if (mask & SDHCI_RESET_ALL)
> +		sg2042_sdhci_phy_init(host);
> +}
> +
>  static const struct sdhci_ops sdhci_dwcmshc_ops = {
>  	.set_clock		= sdhci_set_clock,
>  	.set_bus_width		= sdhci_set_bus_width,
> @@ -728,6 +808,16 @@ static const struct sdhci_ops sdhci_dwcmshc_cv18xx_ops = {
>  	.adma_write_desc	= dwcmshc_adma_write_desc,
>  };
>  
> +static const struct sdhci_ops sdhci_dwcmshc_sg2042_ops = {
> +	.set_clock              = sdhci_set_clock,
> +	.set_bus_width          = sdhci_set_bus_width,
> +	.set_uhs_signaling      = dwcmshc_set_uhs_signaling,
> +	.get_max_clock		= dwcmshc_get_max_clock,
> +	.platform_execute_tuning = th1520_execute_tuning,
> +	.reset                  = sg2042_sdhci_reset,
> +	.adma_write_desc        = dwcmshc_adma_write_desc,
> +};
> +
>  static const struct sdhci_pltfm_data sdhci_dwcmshc_pdata = {
>  	.ops = &sdhci_dwcmshc_ops,
>  	.quirks = SDHCI_QUIRK_CAP_CLOCK_BASE_BROKEN,
> @@ -763,6 +853,13 @@ static const struct sdhci_pltfm_data sdhci_dwcmshc_cv18xx_pdata = {
>  	.quirks2 = SDHCI_QUIRK2_PRESET_VALUE_BROKEN,
>  };
>  
> +static const struct sdhci_pltfm_data sdhci_dwcmshc_sg2042_pdata = {
> +	.ops = &sdhci_dwcmshc_sg2042_ops,
> +	.quirks = SDHCI_QUIRK_CAP_CLOCK_BASE_BROKEN |
> +		  SDHCI_QUIRK_INVERTED_WRITE_PROTECT,

is "wp-inverted" property better?

> +	.quirks2 = SDHCI_QUIRK2_PRESET_VALUE_BROKEN,
> +};
> +
>  static void dwcmshc_rk35xx_postinit(struct sdhci_host *host, struct dwcmshc_priv *dwc_priv)
>  {
>  	/*
> @@ -882,6 +979,58 @@ static int dwcmshc_th1520_init(struct device *dev,
>  	return 0;
>  }
>  
> +static int dwcmshc_sg2042_clks_enable(struct dwcmshc_priv *dwc_priv)
> +{
> +	int ret = 0;
> +	struct sg2042_priv *soc = dwc_priv->priv;
> +
> +	if (soc)
> +		ret = clk_bulk_prepare_enable(SG2042_MAX_CLKS, soc->clks);
> +	return ret;
> +}
> +
> +static void dwcmshc_sg2042_clks_disable(struct dwcmshc_priv *dwc_priv)
> +{
> +	struct sg2042_priv *soc = dwc_priv->priv;
> +
> +	if (soc)
> +		clk_bulk_disable_unprepare(SG2042_MAX_CLKS,
> +					   soc->clks);
> +}
> +
> +static int dwcmshc_sg2042_init(struct device *dev,
> +			       struct sdhci_host *host,
> +			       struct dwcmshc_priv *dwc_priv)
> +{
> +	int err;
> +	struct sg2042_priv *soc = NULL;
> +
> +	soc = devm_kzalloc(dev, sizeof(struct sg2042_priv), GFP_KERNEL);
> +	if (!soc)
> +		return -ENOMEM;
> +
> +	soc->clks[0].id = "card";
> +	soc->clks[1].id = "timer";

Interesting, only "card" and "timer", so which clk is for clk input of the ip?

> +	err = devm_clk_bulk_get_optional(mmc_dev(host->mmc), SG2042_MAX_CLKS,
> +					 soc->clks);
> +	if (err) {
> +		dev_err(mmc_dev(host->mmc), "failed to get clocks %d\n", err);
> +		return err;
> +	}
> +
> +	err = clk_bulk_prepare_enable(SG2042_MAX_CLKS, soc->clks);
> +	if (err) {
> +		dev_err(mmc_dev(host->mmc), "failed to enable clocks %d\n", err);
> +		return err;
> +	}
> +
> +	dwc_priv->priv = soc;
> +	dwc_priv->soc_clks_enable = dwcmshc_sg2042_clks_enable;
> +	dwc_priv->soc_clks_disable = dwcmshc_sg2042_clks_disable;
> +
> +	return 0;
> +}
> +
>  static const struct of_device_id sdhci_dwcmshc_dt_ids[] = {
>  	{
>  		.compatible = "rockchip,rk3588-dwcmshc",
> @@ -907,6 +1056,10 @@ static const struct of_device_id sdhci_dwcmshc_dt_ids[] = {
>  		.compatible = "thead,th1520-dwcmshc",
>  		.data = &sdhci_dwcmshc_th1520_pdata,
>  	},
> +	{
> +		.compatible = "sophgo,sg2042-dwcmshc",
> +		.data = &sdhci_dwcmshc_sg2042_pdata,
> +	},
>  	{},
>  };
>  MODULE_DEVICE_TABLE(of, sdhci_dwcmshc_dt_ids);
> @@ -994,6 +1147,12 @@ static int dwcmshc_probe(struct platform_device *pdev)
>  			goto err_clk;
>  	}
>  
> +	if (pltfm_data == &sdhci_dwcmshc_sg2042_pdata) {
> +		err = dwcmshc_sg2042_init(dev, host, priv);
> +		if (err)
> +			goto err_clk;
> +	}
> +
>  #ifdef CONFIG_ACPI
>  	if (pltfm_data == &sdhci_dwcmshc_bf3_pdata)
>  		sdhci_enable_v4_mode(host);
> -- 
> 2.25.1
>
Chen Wang April 16, 2024, 11:53 p.m. UTC | #2
On 2024/4/16 23:37, Jisheng Zhang wrote:
> On Tue, Apr 16, 2024 at 05:50:57PM +0800, Chen Wang wrote:
[......]
>>   
>> +#define SG2042_MAX_CLKS 2
> I don't think "bulk" is suitable here for max 2 clks, no?
Without "bulk",  I have to prepare/disable two times for each of clocks 
and handle the exception if first one failed etc. I learn this from 
rockchip, it has 3. Why you think it is  not suitable, please advise me, 
thanks.
>> +struct sg2042_priv {
>> +	struct clk_bulk_data clks[SG2042_MAX_CLKS];
> useless either

Sorry, what's this mean?

[......]

>> +
>> +	/* Set ATDL_CNFG, tuning clk not use for init */
>> +	val = FIELD_PREP(PHY_ATDL_CNFG_INPSEL_MASK, 2);
> magic "2" needs a meaningful macro definition.

Agree, will improve this in next version.

[......]

>>   
>> +static const struct sdhci_pltfm_data sdhci_dwcmshc_sg2042_pdata = {
>> +	.ops = &sdhci_dwcmshc_sg2042_ops,
>> +	.quirks = SDHCI_QUIRK_CAP_CLOCK_BASE_BROKEN |
>> +		  SDHCI_QUIRK_INVERTED_WRITE_PROTECT,
> is "wp-inverted" property better?

Yes, l will use this in next revision, thanks.

[......]

>> +
>> +static int dwcmshc_sg2042_init(struct device *dev,
>> +			       struct sdhci_host *host,
>> +			       struct dwcmshc_priv *dwc_priv)
>> +{
>> +	int err;
>> +	struct sg2042_priv *soc = NULL;
>> +
>> +	soc = devm_kzalloc(dev, sizeof(struct sg2042_priv), GFP_KERNEL);
>> +	if (!soc)
>> +		return -ENOMEM;
>> +
>> +	soc->clks[0].id = "card";
>> +	soc->clks[1].id = "timer";
> Interesting, only "card" and "timer", so which clk is for clk input of the ip?

Copy my comments from bindings patch here for quick reference:

 > SG2042 defines 3 clocks for SD/eMMC controllers.

 >- AXI_EMMC/AXI_SD for aclk/hclk(Bus interface clocks in DWC_mshc)
 >  and bclk(Core Base Clock in DWC_mshc), these 3 clocks share one
 >  source, so reuse existing "core".
 >- 100K_EMMC/100K_SD for cqetmclk(Timer clocks in DWC_mshc), so reuse
 >  existing "timer" which was added for rockchip specified.

 >- EMMC_100M/SD_100M for cclk(Card clocks in DWC_mshc), add new "card".

What you meant "clk input of the ip" is "core", right?

[......]